it's an annotation that means, when we're doing a tinderbox UNIVERSE build, don't bother building it

meena: ok, added

still missing i386, this is 15-CURRENT, not 16-CURRENT :P

like i said i'm ignoring non-amd64 for now

that can happen later

i386 will still work fine with this change (i hope)

aye, then I'm not gonna suggest to go track down all other KERNCONFs where that DEBUG stuff is duplicated

we can dedup that later

but explain the PR description that the focus is on amd64 for now until the kinks are ironed out

done

How do I figure out what is the wifi chip I have ? The mini-pc I have came with some USB based wifi-dongle on the back. I've only used the hardwire connection. I am thinking of moving the system into the basement, so would need wifi.

mns: 'pciconf -lv' might help

that will list all PCI devices and what driver (if any) is attached

there we go: options RACCT / options RCTL needs to be added to MINIMAL (not today)

and I think you mentioned: device cpufreq needs to go; as this is a loadable module

OTOH, device cd needs to be added, if we're going to use it as the default kernel when building ISOs

ok biab, need to bisect whatever broke my snd_uaudio

i'm looking for a guide or howto for running netbsd10 guest in bhyve. apparently grub2-bhyve isn't needed. it needs to run without the vm-bhyve tool

f451: see how vm-bhyve does it and run the bhyve command from the log

it's not possible to run that here

it doesn't have to be on the same system

there are already several vms

ah icwym

vm-bhyve seems to use knetbsd

yes

that is grub2-bhyve isnt it

i don't think so? (but i havn't used netbsd since the 3-4 days)

hrm, i guess it is

yeah

/usr/ports/sysutils/grub2-bhyve/work/grub2-bhyve-0.40/grub-core/tests/boot/knetbsd.cfg

also vm-bhyve uses grub

lw, for how long are you available?

jbo: probably about 30 years, then i expect i'll die of something or other

lw, how about tonight? do you happen to be still around in 30-ish minutes (C++)?

jbo: yes, i just need to complain on -current about a broken commit then i'm all yours

jbo: ok, complaining done, what do you need?

jbo: speak ur piece now or i'm going to afk again to rack this switch

lw, sorry I messed up - still at the office. got side tracked

apologies

heading home now, let me know if you're still up in 20min

totally fine if not, no rush

lw, I'd be "ready" now in case you're still up and running

jbo: not sure for how long, but sure

lw, godbolt.org/z/3K93sz48W

Title: Compiler Explorer

1/ why does this not compile?

2/ How would I go about making the stuff at line 27 work based on the non-type template parameter Num (i.e. make it work with values other than 4?

C++20

jbo: godbolt.org/z/493a5jYYr - the array thing is weird, but you want the initialiser to be { { a, b }, { c, d} } so you need std::array({{a, b}, {c, d}})

Title: Compiler Explorer

lw, so that for() can be constexpr?

i expect so yes, but i've never really done any constexpr stuff

that's similar to how I have not really done any ranges stuff yet :D

lw, like I have had to write some shit code yesterday

lw, pastebin.com/Xn73muxN

Title: [[nodiscard]] std::deque<vertex> vertices_from_descrip - Pastebin.com

and I was wondering whether taht can be done with ranges these days

btw, the \ in your original code are unnecessary, you never need \ in a context like that

could do an std::transform too

\ only necessary in macro definitions and string literals

lw, oh, that is correct - didn't pick up on that yet. thanks!

std::deque<vertex> v = vertices | std::views::transform([](auto const &vv) { return m_graph[v]; }) | std::to<std::deque>(); // or something... i can't remember if to<deque> is a thing that exists

s/m_graph[v]/m_graph[vv]/

or just std::deque<vertex> v; std::ranges::transform(vertices, std::back_inserter(v), [](auto const &v) { return m_graph[v]; }); // this may be more readable tbh

ah, so the exact same approach I'd have used with std::transform

but then the ranges version

yes, for every std::algorithm there's an std::ranges::algorithm that takes a range instead of an iterator pair

I need to read up on what a range truly is. conceptionally it's just two iterators, right?

or technically, an iterator and a sentinel?

basically yes, but ranges have more semantics, for example a range can be 'borrowed' which means it's created from a temporary

that is different than the notion of a view, right?

yes, a range is only a view if it's explicitly defined as a view using enable_view

the standard imposes certain requirements on views which you are supposed to follow if you make your range a view, but it's not technically required

I'll allocate some time in the weekend to get a primer on views

seems much more readable than managing those iterators "manually"

i always hated iterators, with ranges and generators, they're much nicer to use now

yeah, I only really got into C++ once C++17 became a thing

that is when I truly "fell in love" with the language

I remember the pre C++11 pain all to well. it was basically like writing C code with some (albeit nice) extra semantics

lw, btw, I am still considering to follow your advice of switching my message payload from std::array<std::uint8_t> to std::array<std::byte>

i started with pre-standard C++, it's remarkable how far the language has come since then... but i have to say, the C++98 standard deserves a *lot* of credit for laying out the basic mechanisms that everything else was built on, even if it was missing a few things

yeah, it seems to have been done in quite a future-proof way

lw, I'm also considering to change that std::uint32_t in my CAN message type over to std::bitset

including the STL in the standard library, and introducing the improved template semantics that never existed before

wasn't c++98 the first one?

johnjaye: C++98 was the first standard, C++ as a language existed for ~15 years prior to that

jbo, I foresee some pain using std::bitset but I need to represent exaclty 29 bits :D

i read a statement that due to operator overloading a = b in c++ could mean anything. lol

but API i'll be interfacing will use uint32_t so will probably have to use std::bitset::to_ulong

johnjaye: that can mean anything in C too. "#define b 47"

that ^

the b yes. can you #define = 47 though

#define uint32_t char for extra fun

this is not an issue in real code if your programmers are not insane

lw, have you worked with std::bitset before?

still i think c++ is probably the best there is. maybe D is better. but everything else is super overengineered

jbo: i've used it once or twice but i wouldn't call myself a bitset expert

D is very limited outside of system programming

lw, I am shocked - how dare you to dissapoint me

lw, -1 commit for you :p

(very much kidding, in case that is not obvious)

the wiki article makes it sound like it has every feature you would ever need

jbo: if you're doing CANbus you're working at a much lower level than i normally deal with

except generics maybe

C has generics :> *duck-and-hide*

(it really does)

generics? you mean _Generics

_Because _This _Is _How _We _Spell _Everything _In _C

i forgot already. you're both committers, right?

lw, yeah I usually deal with very low level stuff. so far I have been doing C++ mostly on application/server level stuff and my low-level stuff is all in C but I'm trying to transition to C++ there. hence the questions, sorry

_Ill _Take _What _Is _A _Namespace _For _Ten _Dollars _Please _Alex

johnjaye: i am not a committer

lw, yes en.cppreference.com/w/c/language/generic

Title: Generic selection (since C11) - cppreference.com

ok. i was wondering how much c/c++ you need to know to maintain a port

zero

none

negative

but wouldn't you need to know something to make a patch to fix a bug?

how is knowing C going to help you fix a bug in a python port? or a ruby port, or a golang port

I have been doing C++ for 15 years and I still don't know any of it so the "how much C++ do you need to know?" question is usually either trick question or simply misplaced

i see

it's also not *really* the maintainer's job to fix bugs in a port, unless the port is really specific to freebsd, but even then you could just open an upstream bug for it

s/the port/the bug

lw, careful now :D

that doesn't make sense to me. if the maintainer doesn't fix the bug who does

jbo: i don't think it's controversial to say that maintainers are not expected to fix any random bug in a port

there's a difference between a port bug and an upstream bug

that would require every maintainer to be an upstreamer developer for their port

lw, I'm not disagreeing

meaning it's not considered freebsd's responsibility to fix other people's bugs?

every committer is expected to fix all chromium and windows bugs - no questions about that.

johnjaye: right. the port should basically work, if a bug prevents it from starting or being useful then... there's no point porting it until that's fixed

johnjaye: but if the port just has an ordinary bug that's something that can be handled upstream, all software has bugs like that

jbo: i can't tell if that's a joke!

johnjaye, neither can I

johnjaye, (it is)

heh

I said it before and I'm going to say it again....

Title: files « chromium « www - ports - FreeBSD ports tree

lw: the way i was imagining it yes every maintainer would be an upstream developer or could be for a port

johnjaye, how do you expect that to work with ports where upstream is not open source?

johnjaye: that's certainly possible but it's not a requirement to be a maintainer

i didn't know freebsd had those

plenty of those

in linux land i think they try to sentence those type of things to ppa's or something.

is there a "chrome" port?

I literally just linked to it :D

PPA is something specific to Ubuntu, it's just a way for people to build packages in an automated way that users can optionally install

freebsd pkg technically also has PPA

"technically" (splitting hairs)

you can have overlays and more than one binary repo

jbo: technically it does not because PPA is a specific terminology of Canonical :-)

lw :D

but also, PPAs are binary repositories, freebsd has no way for random people to provide their own binary packages

this could even go down the path of the differences on how openbsd, netbsd, freebsd, dragonbsd handle this same scenario, no?

aside from building them themselves and setting up their own web server

lw, I diagree. I have my own binary repository that others are using

jbo: i could build my own debian repository too but that's not a ppa. PPAs are hosted by the OS vendor

lw, PPAs are hosted by the OS vendor?!

are you sure about that? I don't think so

plenty of rogue PPAs out there

well i don't know about terminology, it's just in my experience when you want to bypass the GPL and install proprietary software it's usually you download the binary or use a PPA

lw, docs.docker.com/engine/install/ubuntu/#install-using-the-repository

Title: Install Docker Engine on Ubuntu | Docker Docs

jbo: they're all hosted on Launchpad afaik

wat?

seriously?

yup

Title: Personal Package Archives : Ubuntu

jbo: that docker installation is not a ppa, it's just a third-party apt repository

lw, I see. I might be misunderstanding PPA as a term/tech then

> Personal Package Archives (PPA) allow you to upload Ubuntu source packages to be built and published as an apt repository by Launchpad.

yeah I see - sorry for the noise then :D

jbo: "PPA" is a specific thing created by Canonical (the company) for their product Ubuntu Linux, which lets people create hosted apt repositories. it's one particular way of creating an apt repository, but every apt repository is not a PPA

lw, thanks for the info - didn't know!

I thought PPA == 3rd party apt repo

so pretty much all of my statements in this discussion were incorrect then

do bsd's do anything like that?

jbo: shame on you!!

:D

yeah canonical just hosts all that stuff on launchpad it seems

lw, that set_bits() function we looked at earlier... I don't really like that solution with passing in two arrays. How would you go about that? tuple?

bloodyhell, why can't ncurses just use standard section nomenclature.. what the heck is this 3X and 1M garbage

voy4g3r2: those are System V Release 4 manpage categories

voy4g3r2: since ncurses is an implementation of SVr4 curses, it probably borrowed the manpage categories too

is that why some manpages end with funny letters like abc or def3

lw: i see that in my research, i am just like.. why not just "fix" it

johnjaye: that happens with some lua bindings.. i default it too.. if it is a "contrib" package you are beholden to the upstream owner

voy4g3r2: well it's not broken. ncurses is contrib software, those are the correct categories on (for example) Solaris, just not on FreeBSD.

voy4g3r2: i said the other day jokingly half of software is figuring out random software conventions from the 80s

johnjaye: facts

|status: One or more devices has experienced an error resulting in data

| corruption. Applications may be affected.

i have come to the conclusion, the number of manual page "issues" are related to how other projects document things

oh great this again

i am really extremely annoyed that openzfs merged encryption and told people to use it and never mentioned that it doesn't effing work

does it work on linux? that is a common answer i see "It works on linux, it is good enough for us"

voy4g3r2: no it's broken everywhere

it is crazy how linux centric that whole project is

probably my bias, from my frustrations

voy4g3r2: OpenZFS is the merger of Illumos ZFS and ZFSonLinux and, well, it's fair to say ZFSonLinux probably had more users/developers...

i wish we could go back 20 years and shoot Oracle before they bought Sun and then have Sun be the upstream of ZFS

yeah, i get that.. which brings back to the whole PPA/upstream/downstream thing

who is responsible for "fixing" things.. the upstream or the downstream project

amen

Sun's software engineering process was probably one of the best in the entire industry

i always say.. oracle should of just stuck with databases

i wonder how discussions go "My database is slow and i am going to blame the system administrator"

with the whole discussion being internally to the SAME company

voy4g3r2: INCREASE SGA SIZE FOR GREAT PERFORMANCE

we can sprinkle in some network latency for some good measure

haha

you ever get .. stop auditing so much to decrease writes to the database

when i say no to that, they then bring that one in

i once made the SGA on our test server so small the database wouldn't start at all, that was fun to fix, had to learn about startup pfiles or whatever they're called

haha

i have a had few metalink (yes that is how long ago i worked with oracle databases) that say.. enable logging and reproduce the issue

mind you this is with their application services and database

you enable debugging to get some more details of how they are interacting.. the problem "goes away"

they then respond.. there is no issue..

turn of the debugging.. oh look at that.. it appears again

and their UTF-8 support was HORRIBLE back in the day

database would story in UTF-8 and NOT mention Byte Order Mark.. their XML subsystem (in application server) just assumed it was little endian

so when you tried to to eastern european languages.. it would be rejected by the external oracle database..

voy4g3r2: turn off every log. you have no chance to debug make your time

because those numbnuts.. NEVER thought you could have an OPTIONAL byte order mark.. impact a UTF-8 export/import

they would NEVER admit, i was right..

wait why would UTF-8 need a BOM? UTF-8 is byte order independent

lw: i knew there was a lot of love for Sun but I didn't meet a person like that until now

johnjaye: a person like what?

lw: yes it is

someone who wishes oracle didn't buy sun

but when you have a database encoded in one character set and you export to another it does not wor

i never really knew Sun but I've read a lot about it

Title: Byte order mark - Wikipedia

johnjaye: literally everyone wishes that though?

idk. maybe?

lw: root cause would come be switching languages in one file

i actually used a sun workstation so that's how i know about them

lw, thanks again for your help mate!

so some of the file was in american english, and some countries had a requirement to translate to their local country language for submissions

if the language was ALL the same.. no issues, the minute you mixed 2 languages.. the byte order mark was required for the successful exporting of contents

voy4g3r2: that doesn't sound right at all. the entire point of Unicode is it can represent every script, you don't need a special marker to switch scripts

BOM use is optional. Its presence interferes with the use of UTF-8 by software that does not expect non-ASCII bytes at the start of a file but that could otherwise handle the text stream.

(except for some ambiguous CJK characters... which was a mistake on the part of the standard, but i thought they fixed that)

lw: this was 2004-2008 era

so they may have fixed this since then

this was true even in 2004, at least for latin scripts (including eastern european)

and cyrillic

i wish i was making this stuff up :)

i spent months with them on it

sounds like whatever was ingesting the data was broken, but not the database's fault

bfarm.de/EN/Home/_node.html <---them

Title: BfArM - Homepage

they handled eastern european country reporting

the best was, the destination was the same system i was using on my side

it went application -> oracle database -> XML generation subsytem -> Cyclone Gateway (client) -> cyclone gateway (health authority) -> application -> xml subsystem (throw-up) -> oracle database

so they finally put in the optional byte order mark at the beginning of the generated file and it all worked

which comes full circle.. oracle should just stick to databases

tangent done

oracle database character encoding (source): en.wikipedia.org/wiki/ISO/IEC_8859-1

Title: ISO/IEC 8859-1 - Wikipedia

i knew there was more to it, i am old

lw: voy4g3r2: regarding zfs encryption, over on the linux side there are people claiming that it works fine on freebsd (but not on linux)! so much misinformation.

ZFS on linux and FreeBSD is the same ( since FreeBSD 13 )

From my understanding, ZFS on Linux has a worse implementation than the ones on FreeBSD and Illumos. But then again, most of the Linux users don't really care about ZFS.

it is literally the same implementation, same code

what you say might have been true once, but these days it is all one codebase so we share features and bugs with the linux world :)

tykling: One big difference - OpenZFS on FreeBSD doesn't have to content with malicious spiking of the project in the form of every-more-common "GPL-only" kernel symbols.

contend

sure

tykling: the issue with ZoL is that not a single distribution makes it easy to set up a *sensible* (emphasis on this) ZFS root. this is really annoying.

mason: I was just talking to a friend about this today, it is almost movie-worthy level comicbook villainy evil behaviour

tykling: Yep.

what's sensible to dstolfa ?

xLXGHTNXNGx: He can answer for himself, but it's a bit tricky to have /boot on ZFS.

If you do it by hand, you can lay out a root that's largely identical to what FreeBSD does, but getting there is manual work, whereas FreeBSD sets it up with near-trivial ease.

and in case of UEFI you simply can't

angry_vincent: I use ZFS root on Linux under UEFI, but I keep /boot off of ZFS. Alternately you can keep a kernel and initramfs in your ESP and that works too. But both have important data outside of ZFS.

my experience with zfsbootmenu and voidlinux is that the vestigial /boot is on ZFS, and it seems to be essentially automatic. However, ZFSBootMenu works by encasing a Linux kernel and initrd in the .efi bootloader, which kernel then.. kexec's the kernel on /boot on the chosen bootenv While it's automatic, it's a kluge..

±punctuation

Yeah, it's a more hands-off kernel and initramfs in your ESP.

Better to just have the distribution kernel and initramfs in there directly. Far fewer moving parts. FreeBSD does this better than anyone else and has for a while.

True.

Although FreeBSD has a bootloader in ESP.

ZBM just uses the Linux kernal itself as a bootloader.

i'm not sure why my irc client didn't ping me with xLXGHTNXNGx's message, but yes it's the UEFI boot thing. not that it's impossible, it's just a lot of messing around that shouldn't really be necessary

dstolfa: might be beause I didn't start off with your nick

at least that's the thing that annoyed me most. other than that, i have had ubuntu ZFS for a while now on 40T RAIDZ1 which works fine, but that's not rootfs or /boot

xLXGHTNXNGx: that's probably it yeah

i just use default irssi, might be why

yeah, irssi only pings on leading with the nick

at least by default

i had to apply olive oil to my keyboard again

Olive oil?

Yes.

Lubricating mechanical switches, or for flavour?

I have to occasionally apply vaseline to my trackball. It works.

what kind of skill izzue do I suffer from, if I tought it would be possible to lookup the IP from the hostname associated with a jail.

mason: NOTA. Lubricating a stabilization structure.

I can imagine olive oil drying out in time, but I imagine it's way better environmentally than Krytox.

The problem with natural oils like olive oil is that they will 1) go rancid over time 2) adhere dirt, both of which are less desirable. I would suggest a synthetic lubricant, one recommended for locks would be best.

If talking about the environment I can't imagine the amount of synth lub needed on a keyboard would be significant compared to the harm of the keyboard itself. Better to keep the keyboard in use as long as possible now that it is manufactured than to need to scrap the entire thing and start again with a newly manufactored keyboard.

Good point.

My previous long time keyboard was a Mouse Systems (anyone remember them?) keyboard with a longer key stroke and the plastic of the keys would get dry and somewhat bind up in when trying to push them down. It was a keyboard that needed some lube every so often on problematic keys.

But honestly I can't imagine using anything on a trackball as that is as surface I would be touching with my fingers. But I guess there are internal bearings on the support wheels with a trackball so I guess there is that. I have never been a trackball person.

I use a very small amount of vaseline on my trackball which does not get on my fingers. I laughed at Kensington's video which suggests it but it does help.

TB's now have friction mounts which are cheap artificial rubies.

rwp: I was in fact pondering getting another keyboard, but your note about the consumption of resources at manufacture has stilled my greed.

topcat001, Interesting tidbit about the trackball mounts. I did not know they were doing that now.

mason, I was not intending that reaction! Sorry! I was reacting to the comment that olive oil for lube was better for the environment than a synth lube but the amounts used on a keyboard would be miniscule.

If you were to pareto out the millions of gallons/liters of oil the military uses all over the world you would find that our computer use of these resources is insignificant by comparison. If a keyboard is giving me repetitive-stress-injury and I can live healthier with happy fingers and wrists then by all means improve your life and get a better ergonomic keyboard for your health!

rwp: My current keyboard is sufficient. It would have been an unnecessary purchase. :)

hello

mason, There is an old song that I think your response fits. Nobody Likes Me, ... I think I'll go eat worms. It's just a silly song that sticks in your head.

If it makes you happy to have a better keyboard then I think you should not avoid doing it because you are worried you are over consuming. I see you online all of the time. I know you spend half your life plugged into the keyboard. If a better keyboard would improve your life then you should have a better keyboard. I love my Thinkpad keyboard. It was well worth it.

Hello tjk. Welcome to the FreeBSD channel. If you have a question or topic about FreeBSD feel free to jump in. I am going back to a $WORK sprint.

Hello rwp. Are you a bot?

:>

I sometimes on good days will pass a Turing test. But it's hard to tell on other days.

rwp: are you a committer, a maintainer, or other?

johnjaye, I am not. I maintain some things in other worlds but in FreeBSD I am just a user here.

oh ok. i'm not even a user. i'm like... idk, an initiate.

All are welcome at the table! :-)

I am sure you are a FreeBSD user. You just haven't arrived there yet.

right now i'm tinkering with ports in a vm until I figure out something better

That sounds like you are already a user to me. :-)

I am getting called away IRL and must run. BBIAB.

cya^

tomorow friday and the weekend is here

johnjaye: it doesn't get any better than ports

so it turns out I'm still getting drm fault errors even with the intel DDX driver. I really don't like how any particular combination of drivers and settings seems to produce these errors so inconsistently, like where it varies from boot to boot

also turns out intel DDX with the SNA AccelMethod produces fault errors without even needing picom to be running

is it better to use gstat -pb to find available disks than camcontrol devlist, since the latter may miss some disks?

I'd love to actually get some kind of additional detail on what is triggering the fault errors. I still don't even know what a fault error is, tbh

like a page fault is all that comes to mind

johnjaye: it has something to do with IRQ handlers evidently

Title: drm-kmod/drivers/gpu/drm/i915/i915_irq.c at master · freebsd/drm-kmod · GitHub

lw

greetings. is there an equivalent to tomoyo linux but for freebsd?

xisop: While there may be forks, the BSDs are their own OS - it isn't like Linux where there's a bunch of distributions sharing the same kernel and differnciating themselves on the userland.

Oh, welp I just made an ass out of myself - that's what I get for assuming.

xisop: docs.freebsd.org/en/books/handbook/mac

Title: Chapter 18. Mandatory Access Control | FreeBSD Documentation Portal

debdrup: that's fine, i'm just trying to figure out if any exist

lol "Vim" diesel

Check the handbook chapter I linked. ;)

debdrup: oh, nice.

There's also mac(9)